Changes of regional meteorology induced by anthropogenic heat 1 and their impacts on air quality in South China

: Anthropogenic heat (AH) emissions from human activities can change the urban 12 circulation and thereby affect the air pollution in and around cities. Based on statistic data, the 13 spatial distribution of AH flux in South China is estimated. With the aid of the WRF/Chem model 14 in which the AH parameterization is developed to incorporate the gridded AH emissions with 15 temporal variation, the simulations for January and July in 2014 are performed over South China. 16 By analyzing the differences between the simulations with and without adding AH, the impact of 17 AH on regional meteorology and air quality are quantified. The results show that the regional 18 annual mean AH fluxes over South China are only 0.87W/m 2 , but the values for the urban areas of 19 the Pearl River Delta (PRD) region can be close to 60 W/m 2 . These AH emissions can 20 significantly change the urban heat island and urban-breeze circulations in the big cities. In the 21 PRD city cluster, 2-m air temperature rises up by 1.1 ℃ in January and over 0.5 ℃ in July, the 22 boundary layer height increases by 120m in January and 90m in July, 10-m wind speed is 23 intensified over 0.35 m/s in January and 0.3 m/s in July, and the accumulative precipitation is 24 enhanced by 20-40% in July. These changes of meteorological conditions can significantly impact 25 the spatial and vertical distributions of air pollutants. Due to the increases of PBLH, surface wind 26 speed and upward vertical movement, the concentrations of primary air pollutants decrease near 27 surface and increase at the upper levels. But the vertical changes of O 3 concentrations show the 28 different patterns in different seasons. The surface O 3 concentrations in big cities increase with 29 maximum values over 2.5ppb in January, while O 3 is reduced at the lower layers and increases at 30 the upper layers above some megacities in July. This phenomenon should be attributed to the facts 31 that the chemical effects can play a significant role in O 3 changes over South China in winter, 32 while the vertical movement can be the dominant effect in some big cities in summer. Adding the 33 gridded AH emissions can better describe the heterogeneous impacts of AH on regional 34 meteorology and air quality, suggesting that more studies on AH should be carried out in the 35 climate and air quality assessments. 2 , which are similar to those in Seoul of Korea (Lee 269 et al., 2009), Toulouse of France (Pigeon et al., 2007), and some US cities (Sailor and Lu, 2004; 270 Fan and Sailor, 2005). The regional highest value occurs in Hong Kong, with the value exceeding 271 100 W/m 2 . This value is comparable to those in the most crowded megacities, such as Shanghai 272 (Xie et al., 2016), Tokyo (Ichinose et al., 1999), London (Hamilton et al. 2009; Iamarino et al. 273 2012), and Singapore (Quah and Roth, 2012). In Nanning and Haikou, the annual mean AH fluxes 274 over the whole administrative district are close to 10 W/m 2 . Our spatial distribution of AH based 275 on the population reflects the economic activities in South China, suggesting that our method is 276 effective and the results are reasonable. These results can be supported by other previous 277 investigations (Flanner, 2009; Chen et al., 2012a; 2014; Xie et al., 2015; Lu et al., 2016). So, our 278 AH data can be used in models to investigate their impacts on urban climate and air quality. 279


Method for estimating anthropogenic heat fluxes 114
The top-down energy inventory method, which predicts AH emissions based on the statistics 115 data of energy consumption, is the most common approach and widely used all over the world 116

WRF/Chem and its configuration 158
The WRF/Chem version 3.5 is applied to investigate the impacts of AH fluxes on regional 159 meteorology and air quality over South China. WRF/Chem is a new generation of air quality   Fig. 4, Fig. 6, Fig. 8, Fig. 9, and Fig. 10.

184
The detailed options for the physical and chemical parameterization schemes used in this 185 study are shown in Table 1 where SH is the total sensible heat flux in a grid. F V and SH V are the fractional coverage and the 226 sensible heat flux of vegetations, respectively. F U and SH U are those of urban surfaces. AH fixed 227 represents the fixed AH value for all urban areas (Chen et al., 2011). With respect to Grd_AH, we 228 modify Eq. 4 by incorporating the inhomogeneous AH data (Q F ) as follow: 229 The gridded AH fluxes in 2014 from Sect. 2.1 (with the grid spacing of about 4km) are 231 re-projected to domain 2 (9km) by the coordinates of each grid. To account for temporal variability,  increasing. The annual mean AH values in the downtown areas are much higher than the regional 260 ones. For instance, the PRD city cluster always has the highest anthropogenic heat emissions in 261 South China. As shown in Table 3  To evaluate the model performance and clarify the better AH parameterization, the modeling 291 results from Fix_AH and Grd_AH are compared with the observation data in two typical months 292 (January and July). Table 4  Fix_AH and Grd_AH slightly overvalue the 10-m wind speed at four sites. In case Fix_AH, the 317 MB for WS 10 is generally around 1m/s in both months, and the RMSE is less than 2.6 m/s in 318 January and around 2m/s in July. However, the predictions are obviously improved in case 319

285
Grd_AH. The MB decreases to 0.4-0.9 m/s in January and 0.4-0.7 m/s in July, and the values of 320 COR also increase from 0.68 (Fix_AH) to 0.74 (Grd_AH) in July. These improvements from 321 Fix_AH to Grd_AH for T 2 , RH 2 and WS 10 predictions suggest that the default value of 322 WRF/Chem for all urban grids overestimates the AH fluxes in these cities, and our gridded AH    Obviously, there are some differences between the two simulations that use different AH 347 parameterizations. These differences are more obvious in and around big cities because the AH are 348 related with the human activities. Moreover, the differences in January are higher than those in 349 July, implying that the adding of AH can arouse more atmospheric disturbances in winter. From 350 this point of view, Grd_AH can better describe the spatial and temporal heterogeneity of the 351 impacts of AH on regional air quality. 352 Above all, the WRF/Chem simulation accounting for the temporal and spatial distribution of 353 AH (Grd_AH) has a relatively good capability in simulating urban climate and air quality over 354 South China. So, the differences between the modeling results from Non_AH and Grd_AH can be 355 used to quantify the impacts of anthropogenic heat on meteorology and air pollution. 356 Fig. 4a-d, Fig. 5a-d, Fig. 6a-b and Fig. 6g-h show the impacts of AH on surface meteorology, 358 which are defined as the monthly-averaged differences of these meteorological factors between 359

Impacts of AH on meteorological conditions 357
Grd_AH and Non_AH (Grd_AH minus Non_AH) at the surface layer over the modeling domain 2. 360  On account that AH and its diurnal variation are added to the sensible heat item in 372 WRF/Chem, the adding of gridded AH fluxes should increase the modeling results of sensible heat 373 fluxes (SHF) over South China. As shown in Fig. 4a and b, the spatial patterns of SHF changes in 374 both January and July are similar to the spatial distribution of AH fluxes presented in Fig. 2f. The 375 significant increments (> 10 W/m 2 ) of SHF over South China usually occur in and around 376 mega-cities. Especially in the PRD city cluster, adding AH can cause SHF to increase by over 50 377 W/m 2 in both January and July. 378 For the 2-m air temperature (T 2 ) over South China, the AH fluxes can increase their values 379 by adding more surface heat into the atmosphere. As presented in Fig. 4c and d, the patterns of the 380 monthly-averaged T 2 changes are similar to those of SHF ( Fig. 4a and b). In the urban areas, the 381 adding of AH can lead to the significant increase of T 2 , which may enhance the Urban Heat Islands. 382 The maximum T 2 changes are usually found in the city centers of the PRD region, with the typical 383 layer along the line AB (shown in Fig. 1b), and illustrate that the increases of air temperature 401 causing by adding AH are mainly confined near the surface around the cities (Guangzhou and 402 Haikou). These changes of air temperature in Guangzhou are more obvious than those in Haikou, 403 because the AH emissions are much higher in Guangzhou. Furthermore, T 2 changes in winter (Fig.  404 4e) are more obvious than those in summer (Fig. 4f), with the monthly mean increment of T over 405 0.7℃ for January while only around 0.4℃ for July in Guangzhou. This phenomenon should be 406 related with the fact that the background heat fluxes are much lower in winter so that the relative 407 increase of T is more obvious. 408

Changes of boundary layer and wind field 409
The warming up of surface air temperature can enhance the vertical air movement in 410 boundary layer (PBL), and thereby can increase the height of boundary layer (PBLH) as well. As 411 shown in Fig. 5a and b, the boundary layer height becomes higher when the AH fluxes are taken 412 into account. The big increments (more than 50m) usually occur in the urban areas of the PRD 413 region. Because relative higher temperature increment in January can induce higher PBL in this 414 cold season, the maximum changing values of PBLH can be 120m for January but only 90m for 415 July.  between Grd_AH and Non_AH. Obviously, the air near the surface of cities becomes dryer. The 440 negative centers occur in the PRD region, the Chao-Shan area, Haikou and Nanning, which are 441 also the AH emission centers occurring in Fig. 2f. In and around these cities, the reductions of 442 surface RH 2 are -3 to -4% in January and -1% to -2% in July. 443 It was reported that the enhanced vertical air movement can transport more moisture from the 444 surface to the upper layer, and thereby can modify the spatial and vertical distributions of moisture 445 (Xie et al., 2016). This effect mechanism can be clearly illustrated by Fig. 6c-f in this study. As 446 shown in Fig. 6c

472
More moisture transported from surface into the mid-troposphere can increase the 473 precipitation in these urban areas as well. Fig. 6g and h illustrate the enhanced rainfall over South 474 China both in January and July. Because of the negligible accumulative precipitation in winter, 475 there are no significant differences between the Grd_AH and Non_AH simulations for rainfall in 476 January. But in July, the increment of rainfall can be more than 50mm in and around big cities. 477 Moreover, according to the dominant southeast wind in summer, the moisture can be transported 478 to the downwind areas of the PRD city cluster, which causes the increases of rainfall in the 479 northwest part of Guangdong province with the maximum value over 80 mm. 480

Diurnal pattern of the changes 481
In order to better understand the different impacts of AH in the daytime and at night, the 482 monthly-averaged diurnal variations of T 2 and PBLH in January and July over the urban areas in 483 Guangzhou are calculated based on the results from Grd_AH and Non_AH. As shown in Fig. 7a  484 and b, adding AH fluxes can lead to an obvious increase of 2-m air temperature in both months, 485 with the daily mean increase of 1.5℃ for January and 0.6 ℃ for July. The increment of T 2 at night 486 in January (1.69℃) is larger than that in the daytime (1.31℃), whereas the changes during the 487 whole day in July are all around 0.6℃, which suggests that AH can weaken the diurnal T 2 488 variation in winter. With respect to PBLH, the AH fluxes can also result in a higher boundary layer. 489 In July (Fig. 7d), the increment of PBLH nearly keeps a constant value of 54m (4.7%) from 490 morning till night. However, in January (Fig. 7c), the nighttime increase of PBLH is much higher 491 than that in the daytime. This phenomenon may be related with the facts that the absolute PBLH 492 values are lower and the air temperatures increase more in the winter nights.

Changes of the spatial and vertical distribution of PM 10 502
Since adding AH changes the meteorology conditions, it can affect the transportation and 503 dispersion of air pollutants as well. Fig. 8a and b show the effects of AH on the spatial distribution 504 of PM 10 at the surface layer over South China in January and July. They illustrate that the 505 concentrations of PM 10 decrease in both season near the big cities, including the PRD city cluster, 506 the Chao-shan area, and Nanning etc. The maximum reductions occur in the PRD region, with the 507 monthly mean value over -10μg/m 3 for January and about -5μg/m 3 for July. Compared with the 508 distribution of AH emissions as well as their effects on meteorological conditions, the main causes 509 resulting in the reduction of surface PM 10 should be attributed to the increase of PBLH, vertical 510 upward air flow and surface wind speed, which can all facilitate PM 10 transport and dispersion 511 within the urban boundary layer. For another, as shown in Fig. 6h, the rainfall around the PRD    Fig. 8c for January, the decreases 537 of PM 10 manly confined at the surface, with the typical reductions over -8μg/m 3 . Meanwhile, there 538 are obvious increases of PM 10 concentrations at the upper levels, with the increments over 2μg/m 3 539 from the 980hPa layer to the 850hPa layer (approximately from 500m to 1500m). But for July (Fig.  540 8d), from the surface to the 850hPa layer over Guangzhou, the PM 10 concentrations are all reduced 541 over -1μg/m 3 , with the maximum values over -4μg/m 3 on the ground. The increasing zones only 542 occur at the upper layers above 1.5km, with the increments over 1μg/m 3 . This significant seasonal 543 difference for the vertical distribution of PM 10 changes over Guangzhou should be related with the 544 fact that the atmosphere is more unstable and convective in summer than in winter, which can be 545 further proven by the phenomenon that the enhanced upward air movement in July is stronger than 546 that in January (shown in Fig. 6e and f). It should be noted that the vertical changes of PM 10 in 547 Haikou are indistinctive, implying that the surface air pollutants cannot be remarkably affected by 548 adding AH if the heat emission fluxes are less than 10 w/m 2 . Furthermore, the low particle 549 pollution level may be another cause for the negligible vertical changes of PM 10 in Haikou. 550 impacts of AH in the daytime and those at night. In July (Fig. 8f), the decreases mainly occur from 555 6:00 to 17:00. In January (Fig. 8e), the decreases are -8.8μg/m 3 from 8:00 to 18:00 and -11.9μg/m 3 556 from 19:00 to 7:00, with the maximum reduction of -36.9μg/m 3 at 21:00. This pattern has a 557 reverse correlation with the changes of PBLH shown in Fig. 7c  for both January and July. In January (Fig. 9a), the maximum O 3 differences occur in the big cities 563 of the PRD region, with the monthly mean increment over 2.5ppb. In July (Fig. 9b)

594
The mechanism how the AH fluxes influence the spatial and vertical distribution of O 3 is 595 more complicated than that for PM 10 . Only taking the physical effects that just impact O 3 transport 596 and dispersion into account, we can merely deduce that O 3 is seemingly reduced at the surface and 597 variation. Finally, the WRF/Chem simulations are performed, and the differences between the 636 cases with and without adding AH are analyzed to quantify the impacts of AH. 637 The results show that high AH fluxes generally occur in and around the cities. In 2014, the There is an important question asked many times by scientists about whether anthropogenic 661 heat emissions contribute to global warming. Although the answers are probably negative, the 662 systematic analyses of AH over South China in this paper can enhance the understanding of the 663 magnitude of AH emission from megacities and its impact on regional meteorology and 664 atmospheric chemistry. Compared with the effects from urban land use (Wang et al., 2007;2009b; 665